AbstractObjectiveTo investigate: (i) how lunch frequency of adolescents varies between schools and between classes within schools; (ii) the associations between frequency of lunch and individual sociodemographic factors and school characteristics; and (iii) if any observed associations between lunch frequency and school characteristics vary by gender and age groups.DesignCross-sectional study in which students and school headmasters completed self-administered questionnaires. Associations were estimated by multilevel multivariate logistic regression.SettingThe Danish arm of the Health Behaviour in School-Aged Children study 2010.SubjectsStudents (n 4922) aged 11, 13 and 15 years attending a random sample of seventy-three schools.ResultsThe school-level and class-level variations in low lunch frequency were small (intraclass correlation coefficient <2·1 %). At the individual level, low lunch frequency was most common among students who were boys, 13- and 15-year-olds, from medium and low family social class, descendants of immigrants, living in a single-parent family and in a reconstructed family. School-level analyses suggested that having access to a canteen at school was associated with low lunch frequency (OR=1·47; 95% CI 1·14, 1·89). Likewise not having an adult present during lunch breaks was associated with low lunch frequency (OR=1·44; 95% CI 1·18, 1·75). Cross-level interactions suggested that these associations differed by age group.ConclusionsLunch frequency among Danish students appears to be largely influenced by sociodemographic factors. Additionally, the presence of an adult during lunch breaks promotes frequent lunch consumption while availability of a canteen may discourage frequent lunch consumption. These findings vary between older and younger students.